"What to actually measure in a team"

The question is almost never whether to measure a team. Companies measure their teams constantly — engagement scores, activity metrics, satisfaction surveys, output totals. The question that actually matters is what to measure, and most team measurement answers it badly, tracking what is easy to count rather than what actually predicts how the team will perform. A company can measure its teams heavily and learn almost nothing useful about them, because the things it is measuring are proxies, results, or aggregates that do not tell it the one thing measurement is for: the state of the things that drive performance.

What to measure in a team is the eight dimensions of team health — trust, communication, alignment, collaboration, decision-making, energy, belonging, and leadership — because those are the underlying causes of how a team performs, and measuring causes is what lets you act before the results arrive. Most team metrics measure something else: the activity, which is how busy the team is; the output, which is a lagging result; or a general engagement score, which blurs many distinct things into one uninformative number. Understanding why the dimensions are the right measurement target, and why the common alternatives are not, is understanding the difference between measurement that informs action and measurement that merely fills a dashboard.

Measure causes, not results

The most important principle in choosing what to measure is to measure causes rather than results. A team's results — its output, its hit targets, its delivery — are lagging indicators: by the time they show, the thing that produced them has already happened, so a result tells you what already occurred rather than what to do now. The causes — the state of the dimensions that produce the results — are what you can act on, because they precede the results and moving them changes what the results will be. Measuring the result tells you the score after the game; measuring the cause tells you the state of the team while there is still time to affect the score.

This is why output metrics, however precise, are poor guides to team health. A team's output reflects its dimensions filtered through everything else that affects output — the market, the resources, the luck — so a good output can hide a deteriorating team and a bad output can come from a healthy team hitting hard conditions. The output is a lagging, noisy result, and reading it as a measure of team health confuses the effect with the cause. Measuring the dimensions directly cuts through that noise: it reads the causes of performance rather than the performance itself, which is both more actionable and more accurate as a picture of the team, because it is not contaminated by all the non-team factors that shape output. Measure the causes, and you can act; measure the results, and you can only react.

Why activity is not health

The most common thing companies measure is activity — meetings held, messages sent, hours logged, tasks moved — because activity is easy to count and produces satisfying dashboards. But activity measures how busy a team is, not how healthy it is, and the two are nearly independent. A team can be intensely busy and deeply unhealthy, spending all its activity pulling in different directions or grinding through dysfunction; and a team can be healthy and less visibly busy, coordinating so well that it does not need the frantic activity a dysfunctional team generates. Activity counts the motion and says nothing about whether the motion is productive, which is exactly what you need to know.

This is why activity metrics are actively misleading as measures of team health. High activity can signal a healthy, engaged team or a team thrashing in dysfunction, and the metric cannot tell them apart, so it provides a number that looks informative and is not. Worse, optimising for activity metrics pushes teams toward more visible motion regardless of whether it helps — more meetings, more messages, more logged tasks — which can degrade the very health the metric was supposed to track. A team that games its activity metrics is a busier, not a healthier, team. The dimensions, by contrast, measure whether the team is actually functioning well beneath the activity, which is the thing activity was being used as a proxy for and is a bad proxy of. Measure the health, not the busyness, because busyness is not health and often disguises its absence.

The trouble with a single engagement score

Many companies reduce team measurement to a single engagement score, and while this is better than activity metrics — it at least aims at something like health — it fails by compressing many distinct things into one number. A general engagement score blends trust, belonging, energy, and several other dimensions into a single figure, so when it moves, you know something changed but not what, and when it is low, you know something is wrong but not which thing. The score tells you there is a problem and withholds the one piece of information you need to act on it: what the problem actually is.

This compression is why a single score cannot guide action, even when it accurately detects that something is off. A low engagement score could reflect a trust problem, a belonging problem, an energy problem, or a leadership problem — each of which requires a completely different response — and the score cannot distinguish them, so it leaves you knowing you must act without knowing how. The eight dimensions separate what the general score blurs, reading each dimension distinctly so that a problem is not just detected but located and named. This is the difference between a smoke alarm and a diagnosis: the score tells you there is smoke, the dimensions tell you where the fire is. A company serious about acting on its measurement needs the dimensions, because the single score detects problems it cannot help you solve.

Specificity is what makes a metric actionable

A good team metric has to be specific enough to act on, and this is where most team measurement fails. A metric that says "engagement is down" is not actionable, because it does not say what to do; a metric that says "the team's decision-making is the weak dimension" is actionable, because it points at a specific thing to address. The value of a metric is not just that it detects a problem but that it locates it specifically enough to guide a response, and specificity is exactly what the aggregate metrics lack and the dimensions provide.

This is why measuring the dimensions separately is worth the additional effort over a single score. Each dimension is specific enough that a reading of it points toward a specific kind of intervention — a trust reading toward safety work, a decision-making reading toward decision work, a belonging reading toward belonging work. The specificity is what connects the measurement to the action, turning a diagnosis into a direction. This is the same logic that runs through the method: a reading is valuable because it aims the intervention, and it can only aim the intervention if it is specific enough to point at a target. A measurement that detects a problem without locating it specifically leaves you knowing you have a problem and not knowing what to do, which is measurement that informs worry rather than action. Specificity is what makes the difference, and the dimensions are specific where the aggregates are not.

Leading indicators buy time

The dimensions are leading indicators, which is the third reason they are the right thing to measure: they move before the visible outcomes do, which means measuring them buys time to act before the outcomes arrive. A dimension deteriorating — trust eroding, belonging fraying, energy depleting — shows in a reading of the dimension well before it shows in the lagging outcomes like attrition, missed targets, or visible dysfunction. Measuring the dimension catches the problem in the window when it is still cheap to address, before it has compounded into the expensive outcome.

This is what separates the dimensions from the lagging metrics companies usually rely on. Attrition, missed targets, and visible breakdowns are real signals, but they are late — by the time they show, the underlying deterioration has already done its damage, and the response is remedial rather than preventive. The dimensions are early, so measuring them lets a company act in the window before the lagging signal, which is the difference between preventing a problem and cleaning up after it. This distinction between leading and lagging signals is worth its own treatment, and we give it one in leading versus lagging signals of team health — but the core point belongs here too: the dimensions are the right thing to measure partly because they are early, and early is what makes measurement preventive rather than merely diagnostic of damage already done.

What the dimensions measure that surveys miss

Standard engagement surveys tend to measure satisfaction — whether people are happy, whether they would recommend the company, how they feel about various aspects of work — and satisfaction is related to team health without being the same thing. A team can be reasonably satisfied and unhealthy in ways satisfaction does not capture, because satisfaction measures how people feel about their situation while the dimensions measure how the team actually functions, and the two can diverge. A team can be satisfied and unable to decide, satisfied and misaligned, satisfied and quietly depleting — satisfaction is not sensitive to the functional problems the dimensions detect.

This is why measuring the dimensions catches things satisfaction surveys miss. The dimensions read the team's actual functioning — whether it can decide, whether it is aligned, whether it collaborates — which is the thing that drives performance, and which satisfaction only loosely tracks. A satisfaction survey can come back fine while a dimension is failing, because people can be reasonably content and the team can still be functionally constrained in a way that will eventually hurt both the performance and, later, the satisfaction. Measuring the dimensions directly reads the functioning rather than the feeling about it, which is both a leading indicator of the feeling and a more direct measure of what actually determines how the team performs. Satisfaction is worth knowing and it is not the same as team health, and measuring it in place of the dimensions measures the feeling while missing the function.

Reading the dimensions is the measurement

Once the dimensions are established as the right target, the measurement is a reading of them — a structured assessment of where each dimension stands for a specific team, of the kind the scan produces. This is not a survey in the usual sense, because it is not aggregating satisfaction into a score; it is reading the state of each dimension distinctly, so that the output is a specific, actionable picture of which dimensions are strong and which are constraining the team. The reading is the measurement, and its value is exactly the specificity and the cause-level focus that the alternatives lack.

And because the dimensions can be read again, the measurement supports the before-and-after comparison that turns a reading into proof, as Day 14, 30 and 60 describes. A single reading locates the constraint; a second reading, after an intervention, shows whether the constraint moved — which is how measuring the dimensions becomes not just diagnosis but evidence of change. This is the full value of measuring the right thing: it aims the intervention going in and proves the result coming out, both of which depend on measuring the causes specifically rather than the results in aggregate. Measure the dimensions, and measurement becomes the spine of the whole method — the aim before and the proof after — rather than a dashboard that detects problems it cannot help you solve.

Measure against the team, not against a benchmark

A tempting mistake in team measurement is to measure a team against an external benchmark — an industry average, a company-wide norm, a target score — and treat the gap from the benchmark as the thing to fix. This imports a standard from outside that may not fit the team, and it turns measurement into comparison rather than diagnosis. A team scoring below a benchmark on some dimension is not necessarily constrained by that dimension; a team scoring above it is not necessarily fine. The benchmark says how the team compares to others, which is a different question from what is actually constraining this team.

The more useful comparison is the team against itself over time. What matters is not whether a team's trust reading is above or below some average, but whether trust is the dimension holding this team back, and whether it is improving or deteriorating. A team's own trajectory — is this dimension moving in the right direction — is more actionable than its position relative to a benchmark, because the trajectory reflects whether what the team is doing is working, while the benchmark reflects only how the team stacks up against a population it may have little in common with. Benchmarks can provide context, but they are not the measurement, and treating the gap from a benchmark as the problem to solve aims the work at matching a population rather than at fixing this team's actual constraint. Measure the team against itself, over time, because that is what tells you whether it is getting healthier — which is the only comparison that reflects whether the work is working.

How often to measure

If the dimensions are the right thing to measure, a natural question is how often, and the answer follows from what measurement is for. Measuring too rarely means problems compound between readings, undetected until the next one, which wastes the leading-indicator advantage that made the dimensions worth measuring — a dimension that deteriorates and recovers between annual readings was never caught in the window when catching it mattered. Measuring too often turns the reading into a burden and can create the surveillance dynamic that measurement should avoid, as measurement without surveillance describes. The right frequency is the one that catches problems in time to act without turning measurement into monitoring.

In practice, this means measuring around the moments that matter and at a sustainable baseline rhythm, rather than either constantly or almost never. A reading before an intervention aims it; readings after it, at intervals like Day 14, 30 and 60, prove and track the change; and a periodic baseline reading catches drift that no specific intervention prompted. The cadence follows the purpose: measure when a reading will inform a decision or verify a change, not on a fixed calendar for its own sake. A team measured thoughtfully, around the moments where the reading matters, gets the leading-indicator benefit without the burden; a team measured constantly gets the burden without proportional benefit, and a team measured almost never loses the early-warning value entirely. Match the frequency to what the measurement is for, and it stays useful rather than becoming either noise or ritual.

Few enough to hold, distinct enough to act

A practical virtue of the eight dimensions as a measurement target is that they are few enough to hold in mind and distinct enough to act on — a balance that most measurement schemes miss in one direction or the other. A single score is easy to hold and useless to act on, because it collapses everything into one number; a sprawling dashboard of dozens of metrics is specific but impossible to hold, so no one reasons across it and it becomes decoration. Eight dimensions sit in the usable middle: a leader can actually keep all eight in view, reason about which is constraining a team, and act on the specific one, without either the over-compression of a single score or the overload of a hundred metrics.

This matters because a measurement that cannot be held is not used, whatever its precision. The point of measuring is to inform judgment, and judgment works with a small enough set of distinct things to reason across — which is what the eight dimensions provide and what both the single score and the sprawling dashboard fail to. Measuring the right things in a form a person can actually hold and act on is part of measuring well, and the dimensions are calibrated to that human limit: enough to capture what matters, few enough to keep in mind, distinct enough to point at action. A measurement scheme that ignores this produces numbers no one reasons with, which is measurement that informs nothing however much it counts.

A worked example

A company measures its teams with a quarterly engagement survey that produces a single score per team, and one team's score drops. The company knows something is wrong and has no idea what, so it responds generically — a morale event, a manager check-in, some general attention — hoping to lift the score. The score does not recover, because the generic response was aimed at nothing specific, and the actual problem, whatever it was, went unaddressed because the measurement never identified it. The company measured heavily and learned only that a problem existed, which was not enough to solve it.

Measuring the dimensions would have located the problem. A reading of the eight dimensions would have shown that the drop was specifically a decision-making failure — the team had become unable to surface and resolve disagreement, and its execution was stalling — which points at a specific intervention rather than a generic one. Same detected problem, but now named and located, so the response can be aimed at the actual constraint and its effect verified in a follow-up reading. The company that measures a single score detects the problem and cannot solve it; the company that measures the dimensions detects it, locates it, acts on it, and proves the result. The measurement was heavier in the first case and useless; lighter in what it asked of people and far more useful in the second, because it measured the right thing.

Measure the causes, specifically, early

What to measure in a team is the eight dimensions of team health, because they are the causes of performance rather than its results, they are specific enough to act on rather than aggregated into an uninformative score, and they are leading indicators that move before the outcomes do. Most team measurement fails on all three counts — measuring activity, which is busyness not health; output, which is a lagging result; or a single engagement score, which detects problems it cannot locate. The dimensions measure the right thing in the right way, which is what makes measurement actionable.

The discipline is to measure causes rather than results, specifically rather than in aggregate, and early rather than late — which the dimensions do and the common alternatives do not. Measure the dimensions and a reading both aims the intervention going in and proves the result coming out, turning measurement from a dashboard that detects problems into the spine of a method that solves them. Measure the causes, specifically, early, and measurement finally does what it was always supposed to do: tell you not just that something is wrong but what, in time to fix it — which is the whole reason to measure a team rather than merely to watch its numbers move.